Research map / Learning theory and algorithms / Kernels and regression theory
Early Stopping for Nonparametric Testing
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use the proposed data-dependent stopping rule T* rather than an iteration count far below or above its optimal order | Functional gradient descent nonparametric test · Polynomial and exponential-polynomial decay kernels | testing power at the minimax separation rate: n^{4m/(4m+1)}; n (\log n)^{-1/(2p)} | Helped |
| Stop far earlier than T* (ablation) | Functional gradient descent nonparametric test · Polynomial and exponential-polynomial decay kernels | testing power at the minimax separation rate: t\ll T*; power \leq \alpha | Hurt |
| Stop far later than T* (ablation) | Functional gradient descent nonparametric test · Polynomial and exponential-polynomial decay kernels | testing power at the minimax separation rate: t\gg T*; power \leq \alpha | Hurt |
| Use stopping exponent gamma = 2/3 instead of the optimal gamma = 1 (ablation) | Early stopping test · Second-order Sobolev kernel (m = 2), simulated testing | power: gamma = 2/3 | Hurt |
| Use stopping exponent gamma = 4/3 instead of the optimal gamma = 1 (ablation) | Early stopping test · Second-order Sobolev kernel (m = 2), simulated testing | power: gamma = 4/3 | Hurt |
| Use stopping exponent gamma = 2/3 instead of the optimal gamma = 1 (ablation) | Early stopping test · Gaussian kernel (p = 2), simulated testing | power: gamma = 2/3 | Hurt |
| Use stopping exponent gamma = 4/3 instead of the optimal gamma = 1 (ablation) | Early stopping test · Gaussian kernel (p = 2), simulated testing | power: gamma = 4/3 | Hurt |
| Use the proposed early stopping method rather than 10-fold cross-validation to select the penalty | Nonparametric testing · Gaussian kernel (p = 2), simulated testing | power | Helped |
| Use the proposed early stopping method as sample size increases | Nonparametric testing · Gaussian kernel (p = 2), simulated testing | power relative to Oracle ES | Helped |
| Increase signal strength c | Nonparametric testing · Second-order Sobolev kernel, simulated testing | power: c=0.5, c=0.8, c=1.0, c=1.2 | Helped |
Closest papers by meaning
- Benefits of Early Stopping in Gradient Descent for Overparameterized Logistic Regression (ICML 2025 · similarity 0.86)
- Minimax Optimal Early-Stopped Gradient Descent for Gaussian Mixture Classification (NeurIPS 2026 · similarity 0.84)
- Optimal Algorithms for Augmented Testing of Discrete Distributions (NeurIPS 2024 · similarity 0.84)
- Learning the Optimal Stopping for Early Classification within Finite Horizons via Sequential Probability Ratio Test (ICLR 2025 · similarity 0.83)
- Kernel-based guarantees for nonlinear parametric models in Bayesian optimization (NeurIPS 2026 · similarity 0.83)
- Kernel Selection is Model Selection: A Unified Complexity-Penalised Approach for MMD Two-Sample Tests (NeurIPS 2026 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.